2011-01-31

WDM Fundamentals

Welcome to the WDM Fundamentals e-Learning Course.

1
 2011-01-31

Objectives
In this e-learning course we will cover the following topics:
› Propagation in Optical Fiber
› Attenuation in Optical Fiber
› Linear Phenomena: CD and PMD
› Non Linear Phenomena: SPM, XPM and FWM
› Stimulated Scattering
› DWDM Amplifier Technology

›On completion of this course the participants will be able to:

1.Describe in detail WDM transmission technology.
2.Understand transmission impairment in optical fibers.
3.Understand the main concepts related to the feasibility of a
WDM link

In this e-learning course we will cover the following topics:

•Propagation in Optical Fiber
•Attenuation in Optical Fiber
•Linear Phenomena, such as Chromatic Dispersion (CD) and Polarization Mode
Dispersion (PMD)
•Non Linear Phenomena, such as Self Phase Modulation (SPM), Cross Phase
Modulation (XPM) and Four Wave Mixing (FWM)
•Stimulated Scattering
•DWDM Amplifier Technology

On completion of this course the participants will be able to:

–Describe in detail WDM transmission technology.
–Understand transmission impairment in optical fibers.
–Understand the main concepts related to the feasibility of a WDM link

2
 2011-01-31

Propagation into
Optical Fibers

Tx Rx

???

Let’s now introduce light propagation into optical fibers: we will discuss the
effects and the impairments we encounter in designing WDM networks.

3
 2011-01-31

THE ELECTROMAGNETIC WAVE SPECTRUM

X RAYS LIGHT RADIO WAVES
RAYS Frequency [Hz]
19 18 17 16 15 14 13 12 11 10
10 10 10 10 10 10 10 10 10 10

-10 -9 -8 -7 -6 -5 -4 -3 -2
10 10 10 10 10 10 10 10 10
Wavelength [nm]

200 300 500 700 900 2,000 3,000 5,000 8,000

100 1,000 10,000
UV Wavelength [nm]
ULTRA
1500 to INFRA RED
VIOLET
1600 nm
VISIBLE 1300 nm
850 nm The WDM spectrum

Let’s start giving a position in terms of frequency to the bandwidth

used for optical fiber transmission.
As known, we can consider three main portions of bandwidth in
the electromagnetic wave spectrum :
•X Rays
•Light Rays
•Radio Waves
The transmission bandwidth for standard optical fiber belongs to
the light rays range.
In this range, standard transmission windows are actually
allocated in what is called the WDM spectrum.
During this course we’ll analyze in details the characteristics of
optical fibers in the WDM spectrum and the reason why this
particular frequency range has been chosen for transmission
purposes will appear clear.

4
 2011-01-31

Attenuation in
Optical Fibers

What is Attenuation?
The first characteristic of optical fibers to be taken into account is the
optical signal attenuation impairment, meaning the fact that the level of the
transmitted optical signal decreases with the covered distance.
Due to physical reasons, each type of fiber presents a different
attenuation pattern, depending on the frequency of the optical signal
propagating into it.
By choosing the best range for optical transmission, we have the
possibility to minimize attenuation impact, therefore increasing the overall
covered distances.

5
 2011-01-31

Attenuation in Fibres
› Light needs to spend energy to propagate
› The attenuation of a fiber section is expressed by the
relation:
Pout = Pin - αL
Pout
α
z
Pin
L
α attenuation constant
L length of the fibre
Pin / Pout optical input/output power

Why does Attenuation of signals happen in optical fibers?

Because light needs to spend energy to propagate into the fiber.
The attenuation of a fiber section is expressed by an exponential decay
relation, connecting the optical signal output power with the input power.
Using a logarithmic unit system, the relation appears as Pout = Pin – αL,
where α is the actual attenuation characteristic of the fibre and L is the
length of the fibre itself.
Pout and Pin are expressed in dBm, L in km while attenuation is
expressed in dB/km units.

6
 2011-01-31

dB and dBm
› The absolute power is expressed in dBm:

dBm = 10log10(P/1mW)
10 dBm = 10 mW 3 dBm = 2 mW
0 dBm = 1 mW

› The relative attenuation of a section of fibre is expressed

in dB.

dB = 10log10(Pout/Pin)
0 dB = no change -10 dB = 10%
-3 dB = 50% -20 dB = 1%

Why do we use dB (decibel) and dBm?

Because we have to deal with quite a large range of optical power absolute levels,
therefore a logarithmic unit is more appropriate than a linear one.
So a useful way to express absolute powers is through a unit called dBm (that is an
absolute value of power measured with reference to a basic level of 1 milliWatt; actually
we should say dBmW, that is dBmilliWatt, but in electronics the use of an abbreviated
dBm expression, even if not right, is generally accepted) .
But it is also useful to express amplifications or attenuations of signals in dB, not
showing in this case the absolute value of power, but just the ratio between power
levels.
Therefore dB is a measure of relative intensity between two signals but does not provide
any information about the absolute intensity of anyone of the two.
Let's now see some examples:
As we said, dBm is an absolute power measure taken against the reference level of
1mW.
So, zero dBm is exactly equal to the reference level, that is 1 mW; a signal of 3 dBm
means an intensity two times higher than 1mW, therefore 2mW, and a 10 dBm signal
shows ten times more power, therefore 10mW.
dB is instead a relative measure, so zero dB means no power level change, -3dB means
half power compared with the original signal, -10 dB means 10% of original power, -
20dB means 1% power level. Note that a 500 W signal is 3 dB lower than a 1000 W
signal, but the same is true for a signal of 1µW with respect to a signal of 2 µW,
therefore dB are not sufficient to know the actual absolute value of power level.

7
 2011-01-31

Fibre Attenuation VALUES

Window Band [nm] α [dB/km]
I 800÷950 2
Attenuation II 1280 ÷ 1350 0.5
[dB/km]
III 1510 ÷ 1600 0.2
10
I
Ra

io n
yle
1 ig

t
hS II

rp
cat

so
te ri n g

Ab
Ultr III
av

d
i ol e

re
t Ab

f ra
0.1 sorp
t i on

In
Fibre impurities
850 1300 1550 λ [nm]
III Window is the only one used for long distance WDM transmission
systems due to lowest attenuation characteristics

Let’s now have a look to the real attenuation values for optical fibers.
Each kilometer the light travels along the fiber causes an attenuation of the
signal, whose amount depends on its frequency.
The dominating absorption effect is Rayleigh scattering, caused by microscopic
inhomogeneities of the index that can scatter a single ray of light in different
directions. This effect becomes less severe with increasing frequencies: the
medium characteristics do not change, but the wave becomes ‘longer’ and
scattering less sensible.
Moreover photons traveling along the fiber can also be absorbed by impurities.
And, in the first part of the spectrum, an important part of the total loss is related
to ultraviolet Absorption.
Above 1700nm light starts to be directly absorbed by the molecules of silica
(infrared absorption).
This leaves three “windows” available for transmission.
First window, between 800nm and 950nm, with a mean attenuation value around
2dB/Km.
Second window between 1280nm and 1350nm with a mean attenuation value
around 0.5dB/Km.
Third window between 1510nm and 1600nm with a mean attenuation value
around 0.2dB/Km, therefore the lowest possible.
So that’s why the third Window is the only one used for long distance WDM
transmission systems.

8
 2011-01-31

Power independent
linear effects

In this part of the course we'll describe Power-independent linear

phenomena which are present in the optical fiber during transmission.

9
 2011-01-31

Power-independent Linear
effects on light propagation

DISPERSION

POLARIZATION
CHROMATIC
MODE
DISPERSION
DISPERSION

We speak about linear effects when their intensity doesn't depend on the power
level of the incident optical signal.
In general Dispersion is a phenomenon that causes temporal widening of the
pulse traveling along the fiber.
Two main kinds of dispersion are defined:
•Chromatic Dispersion, CD,
•Polarization Mode Dispersion, PMD.

10
 2011-01-31

Chromatic Dispersion
(CD)
(Group Velocity Dispersion - GVD)

Chromatic Dispersion, also called Group Velocity Dispersion.

11
 2011-01-31

Chromatic Dispersion
› CD (GVD) acts as a different propagation delay among the
several components of an optical signal
Single Wave Received
Transmitted pulses spreads
pulses in time

ISI

The energy of a narrow pulse is spread out over a

larger time interval → Inter Symbolic Interference (ISI)

D = [ps/(nm km)]

CD, or GVD as we want to call it, acts as a different propagation delay among the
several components of an optical signal.
What's the meaning of this?
Firstly it has to be noted that a signal in optical transmission is not a single
frequency signal, but a modulated optical signal with a mix of optical components
having different frequencies.
Chromatic Dispersion is caused by the fact that the different chromatic
components, meaning frequency components, of the same light pulse travel
along the fiber with different speeds.
The result is that, at the end of the fiber, the different components of the pulse
arrive with delay with respect to each other and the pulse results to be spread on
a longer time interval.
This broadening of pulses introduces an overlapping among the consecutive bits,
named Inter-Symbolic Interference.
The chromatic dispersion grows in a linear way with the distance, so it can be
evaluated by means of a value called “the chromatic dispersion coefficient of the
fiber”, D, expressed in picoseconds per nanometer and kilometer [ps/(nm km)].

12
 2011-01-31

Optical Fibers CHROMATIC Dispersion

D
[ps/(nm km)]
17 [ps/(nm km)]
20 SMF
18
G.6 -G.654
52
16
6 4 [ps/(nm km)]

4 SF
NZ D 0 [ps/(nm km)]
5
2 G.65

0
1530 1540 1550 1560 λ [nm]
DS F
-2 3
G.65

Moreover we have to take into account that different types of fibers do exist.
In particular, as far as WDM applications are concerned, three major types of
single-mode fibre have been developed with different dispersion profiles.
In chronological order:
SMF Standard Single-mode Fibre: designed to provide zero dispersion at 1310
nm. It is also simply known as G.652, from the ITU-T G.652 Recommendation
which defines its characteristics.
Note that also a a special version of G.652 exists, called G.654, specially
developed with the lowest possible attenuation characteristic for very long
applications, like submarine connections.
DSF, Dispersion-shifted Fiber: designed to provide zero dispersion at 1550 nm
(ITU-T G.653)
NZDSF, Non-zero Dispersion-shifted Fibre: designed to provide small dispersion
at 1550 nm in order to reduce non-linear effects on WDM systems (ITU-T G.655)

In the III Window (1500nm range) the typical Chromatic Dispersion values are
respectively:
•17 [ps/(nm km)] for G.652
•Almost 0 [ps/(nm km)] for G.653
•4 [ps/(nm km)] for G.655

13
 2011-01-31

Limits DUE to Chromatic Dispersion

1000
900 2.5 Gb/s
800 10 Gb/s
Distances [km]

700 40 Gb/s
600

G.652
500
400
G.655

300
200
100
0
2 4 6 8 10 15 17 20
Dispersion coefficient [ps/(nm km)]

Chromatic Dispersion actually introduces limitations to the propagation of

the optical signal.
In particular the maximum achievable distance varies as the inverse of the
square of the bit rate of the information transported by the optical signal
itself; therefore chromatic dispersion is troublesome for high bit rates.
A numeric example:
Let's consider three bit rates: 2.5Gb/s, 10Gb/s, 40Gb/s,
In the case of SMF fiber, Dispersion Coefficient of 17ps/(nm-km) at 1550
nm, following distances are achievable according to the bit rate:
Bit rate = 2.5 Gbit/s maximum length ~ 900 km
Bit rate = 10 Gbit/s maximum length ~ 60 km
Bit rate = 40 Gbit/s maximum length ~ 4 km
In case of NZ DSF, Dispersion Coefficient of 4ps/(nm-km) at 1550 nm
Bit rate = 10 Gbit/s maximum length ~ 350 km
Bit rate = 40 Gbit/s maximum length ~ 50 km
Therefore Chromatic Dispersion is negligible at 2.5G, but it becomes a
severe impairment at 10G and a nightmare at 40G.

14
 2011-01-31

Dispersion Compensation
› DCF = Dispersion Compensating Fibre
(with negative dispersion coefficient)

100 km G.xxx
TX DCF-
DCF-100 RX

› DCF modules are usually expressed in equivalent

length of G.xxx fiber.

Fortunately it is actually possible to compensate CD.

To do so two methods are applicable.
The traditional one calls for the use of special DCF, Dispersion
Compensation Fiber modules, that do present negative dispersion
coefficient compensating the positive CD of the fiber. In effect, DCFs are
purely passive modules, made up of a particular kind of fiber, having an
opposite behavior for chromatic dispersion with respect to standard fiber.
Let's now consider a 100Km link.
DCF modules are inserted along the link.
Combining together line fiber CD effect and DCF effect we obtain CD
compensation.
DCF modules are usually expressed in equivalent length of G.xxx fiber,
meaning that a DCF 100 must be employed in a 100 kilometer link.

15
 2011-01-31

Dispersion Compensation Fiber

D
[ps/(nm km)]

20 SM F
18
16 Residual
Dispersion
2
SMF + DCF
0
1530 1540 1550 1560 λ [nm]
-2

-75
-85 DCF
-95
C - BAND

To understand how the compensation works, we have to take into account

the portion of bandwidth where the transmission system is working.
In this range we consider:
the behavior of chromatic dispersion for a generic SMF,
the behavior for Dispersion Compensation Fiber,
and then the overall result.
Since DWDM channels cover a wide range, the compensation cannot be
perfect for the whole range: central wavelengths can be completely
compensated; edge ones, instead, cannot: this is why, with DCF, there is
always what is called a residual dispersion to be taken into account,
referred to as second order Chromatic Dispersion.

16
 2011-01-31

Dispersion-less Networks
Dispersion compensated system Uncompensated system

10G NRZ/RZ
40G DPSK/DQPSK 40G/100G CP QPSK
Network
Simplification
DCM N x DSA N x SSA

Core network - Core network-

1500km 3000km

Dispersion less networks provide:

›Simplified infrastructure
›Lower power consumption
›Simpler planning and quicker deployment
›Smaller spares handling

The second method to compensate chromatic dispersion is the very latest

one, introduced by new advanced DSP, Digital Signal Processing
technology on the receiver side, in combination with multilevel format
modulated transmission and coherent receiver detection.
In dispersion less networks, optical in line dispersion compensation by
means of DCF modules is replaced by DSP at the Receiver side.
It must be noticed that DSP technology itself can also be used to
compensate PMD.
Dispersion-less networks provide:
•Simplified infrastructure, removing at the same time DCMs, and halving
the amplifier variants,
•Lower power consumption, as DCF high introduced attenuation is
avoided, therefore less amplification is necessary,
•Simpler planning and quicker deployment,
•Smaller spares handling.

17
 2011-01-31

Polarization Mode
Dispersion
(PMD)

The PMD, Polarization Mode Dispersion, is the result of a different

behavior of the two polarization components of the light pulse, due to non
accurate symmetry of fiber core.
This is due to manufacturing imperfections, mechanical stress, bending or
vibration, just to list some of the possible causes of this effect.
If the two polarization components are delayed one with respect to the
other, resulting output is the enlarged pulse.

18
 2011-01-31

Light Polarisation
› In a homogenous medium, the two polarisation modes
have the same propagation constant.
› Ideally they don’t give rise to any spreading of the pulse.

But let’s see this effect in more details.

In a transmission medium, a generic pulse moving along the propagation
axis can be divided into two orthogonal polarization components:
•The Horizontal Polarisation Mode
•And the Vertical Polarisation Mode.
In a homogenous medium, the two polarisation modes have the same
propagation constant.
This means that they travel at the same speed and no pulse spreading
occurs.

19
 2011-01-31

Polarization Mode dispersion

”Dispersed”
Optical Signal

In a real fiber, instead, the two components have different speeds: one is called
fast axis, the other one slow axis, and again there will be inter symbol
interference and distance limitation.
The Differential Group Delay (DGD), which represents the difference in speed
between the two polarization components, is a time-variant random variable and
is measured in picoseconds.
It indicates how big the PMD effect is, therefore how much the optical signal is
“dispersed”.

20
 2011-01-31

DGD and PMD coefficient

› The PMD coefficient of a fibre is defined as the

mean value of DGD per square root of fibre
length: PMD coefficient = <DGD> / L1/2

ps / Km1/2
NOMINAL TYPICAL Main source
G.652 0.5 0.02-0.5 external stresses
internal stresses;
G.653 0.5 0.3-1.0 elliptical core
G.655 0.5 < 0.2 internal stresses
elliptical core;
DCF 0.5 0.2-0.5 internal stresses

The PMD coefficient of a fiber is defined as the mean value of DGD per
square root of the fibre length and is measured in picoseconds per square
root of Kilometers.
According to fiber type, there are different values for PMD coefficient to be
considered:
•0.5 as the general nominal value,
but the typical ones are:
•From 0.02 to 0.5 for G.652
•From 0.3 to 1 for G.653
•Less than 0.2 for G.655
•From 0.2 to 0.5 for DCF

21
 2011-01-31

Limits due to PMD

1000
900 2.5 Gb/s
800 10 Gb/s
Distances [km]

ITU Standard
700 40 Gb/s
600
500
400
300
200
100
0
0.1 0.2 0.5 1 1.5 2 2.5 3
1/2
PMD coefficient [ps/km ]

› PMD can be compensated by means of electronic devices

PMD can limit the performance of optical transmission.

PMD is noticeable at high bit rates and is a significant source of
impairment for ultra-long-haul systems.
This is seldom a problem for 10Gb/s long-haul transmission, but it is a
really serious problem for long-haul transmission at 40G and 100G for
which special modulation formats have been studied and implemented.
PMD can be just taken into account during DWDM network design, so to
have the overall value for each link within the acceptable limits of standard
receivers, or, as we already said,
it can be channel based compensated by means of electronic devices
(DSP technology).

22
 2011-01-31

Power dependent non

linear effects

In this part of the course we'll describe Power-dependent non linear

phenomena which are present in the optical fiber during transmission.

23
 2011-01-31

Power-dependent non-linear
effects on propagation
› In optical systems with high power density in the fiber non-
linear effects become relevant

› Self Phase Modulation (SPM)

› Cross Phase Modulation (XPM)
› Four Wave Mixing (FWM)

› Stimulated Scattering:
– Stimulated Brillouin Scattering
– Stimulated Raman Scattering

Power dependent Non Linear effects are due to a physical phenomenon, called
Kerr's effect, producing a variation of the refractive index (a value indicating
speed of wave in material) connected to the field intensity.
In optical systems with high power density, the fiber non-linear effects become
considerable and the relevant consequences are called:
•SPM, Self Phase Modulation
•XPM, Cross Phase Modulation
•FWM, Four Wave Mixing
Actually, in this conditions, photons interact with the medium and new photons
are emitted.
This phenomenon is called stimulated scattering.
There are two types of stimulated scattering:
•Stimulated Brillouin Scattering
•Stimulated Raman Scattering

24
 2011-01-31

Self Phase Modulation

(SPM)

Self-phase modulation (SPM).

25
 2011-01-31

Self Phase Modulation (SPM)

› Spectral bandwidth may increase or decrease during
propagation into the fiber

wavelength time wavelength

time

SPM name is due to the fact that an optical signal propagating inside a nonlinear medium
modulates its own phase.
When an optical pulse is transmitted through a fiber, there is a time-dependent pulse
intensity.
Due to this reason, the incoming optical signal acquires a temporally varying
instantaneous frequency.
The time-dependent phase change caused by Self Phase Modulation is associated with a
modification of the spectrum.
Spectral bandwidth may increase or decrease during propagation into the fiber.
Similarly to CD, SPM leads to spectral modification that can result in Inter-symbolic
Interference.
The nonlinear phase shift increases with the link length.
The time-dependent nature of the optical signal implies that the SPM-induced phase shift
changes with time.
A temporally varying phase for an optical field implies a frequency shift from its central
value.
The time dependence of the frequency shift is referred to as frequency chirp.
The chirp induced by SPM changes in magnitude with the transmitted distance.
In other words, new frequency components are generated continuously as the optical
signal propagates down the fiber.
Self-phase modulation (SPM) leads to spectral broadening of optical pulses.
As the pulses spread, they tend to overlap and are no longer distinguishable by the
receiver.

26
 2011-01-31

Self-Phase Modulation
› For a given intensity, a different power distribution of the
mode results in a different speed of the pulse.
› Different parts of the pulse experience a different effect:
– D < 0 pulse broadening (SPM and CD add)
– D > 0 pulse compression (SPM can compensate Chromatic
Dispersion)

P peak

According to the pulse shape and the power distribution among the components,
the speed of the pulse will change.
In the same way different components of the pulse experience different effects.
Actually, Self Phase Modulation interacts with Chromatic Dispersion affecting the
shape of the optical pulse.
In particular, in case of a negative CD value, SPM effect and CD effects are
added to each other.
In case of positive CD value, the two effects work in opposite way and they
compensate each other.

27
 2011-01-31

Cross Phase
Modulation
(XPM)

Cross-phase modulation (XPM).

28
 2011-01-31

Multi wavelength systems

› XPM effects are quite important for WDM systems

Multi-
wavelength Received pulses
transmitted are shifted in time
pulses

Single Mode Fiber

Cross-phase modulation is the change in the optical phase of a light beam

caused by the interaction with other beams in a nonlinear medium.
Due to its definition this phenomenon is present in WDM systems.
The XPM effects are quite important for WDM systems since the phase of
each optical channel is affected by both the average power and the bit
pattern of all other channels.
The XPM-induced phase shift should not affect system performance if the
CD effects are negligible, this is because optical receivers only respond to
channel powers and ignore all phase changes.
However, chromatic dispersion of optical fibers converts pattern-
dependent phase shift to power fluctuation, affecting the system
performance.
For this reason, XPM is a major source of nonlinear impairment for most
WDM systems.
XPM can be made negligible with adequate spacing between signal
channels.

29
 2011-01-31

Four Wave Mixing

(FWM)

Four-Wave Mixing (FWM).

30
 2011-01-31

FWM effects
› FWM: multiple channels cross-modulate each other’s
› A number of unexpected frequencies are created.

Along the fiber

F1 F2 F3 Frequency

Four-Wave Mixing (FWM) occurs if multiple frequencies propagate together in a

nonlinear medium such as an optical fiber.
When multiple channels are traveling at the same time along a fiber, they interact
with each other and also with the physical medium: this is called cross
modulation.
This interaction causes the generation of pseudo random pulses, meaning that a
number of unexpected frequencies are created.
Let's make an example of three main signals, F1, F2 and F3 along the fiber.
Multiple unexpected signals are created by FWM.
When an unexpected peak falls over a traffic channel, there is an overlapping
between the original signal and the newly created ones.
This overlapping affects the performance for that channel.

31
 2011-01-31

FWM characteristics
› It is impossible to avoid all FWM products when the
number of channels increases over a threshold

NP = (N3 - N2)/2

FWM ~ 1/(CD)2

› FWM is the major issue for DSF fibre G.653 with zero
dispersion in the band of use

FWM cannot be avoided when the number of channels becomes very high
inside the fiber.
The bigger the number of channels the bigger the number of generated
noise pulses according to the formula:
NP = (N3 - N2)/2
Where NP is the number of auto generated noise pulses, and N is the
number of wavelengths inside the fiber.
Moreover, since FWM is inversely proportional to the square of the CD
value, FWM becomes not relevant for the majority of the types of Single
Mode Fibers that do have a relatively high CD value across all the usable
bandwidth.
As an example let's consider the G.655 fiber, indicated as the best fiber for
DWDM: it presents an average Chromatic Dispersion value in the WDM
working bandwidth higher than zero, even if lower than the value of a
standard SMF.
Well, this little value of chromatic dispersion is still sufficient to contrast the
Four Wave Mixing phenomenon.
But, for DSF (G.653) fibers, at frequencies around the zero CD point, the
phenomenon cannot be neglected, because the very low value of
chromatic dispersion causes important FWM effects.

32
 2011-01-31

Unequal Channel Spacing

› An unequal channel spacing can be adopted in case of low

number of channels

192.7

193.4

193.6

193.8
192.6

192.8
192.9
193.0
193.1
193.2
193.3

193.5

193.7

193.9
194.0
192.5

In case of foreseen strong FWM effects, a solution sometimes could

consist in using an unequal spaced channel grid so to avoid the FWM
effect.
This could be applicable when not all the channels on the grid are
necessary according to network design.

33
 2011-01-31

Stimulated Scattering

As already said, when light propagates through a medium, the photons

interact with silica.
Scattering is a general physical process whereby some forms of radiation,
such as light, are forced to deviate from a straight trajectory by one or
more localized non-uniformities in the medium they go through.
There are two types of scattering:
•elastic scattering, in which the energy (and therefore the frequency) of
the incident particle is preserved and only its direction of propagation is
modified;
•inelastic scattering, in which the kinetic energy of the incident particle is
not preserved.

34
 2011-01-31

Elastic Scattering
In Elastic Scattering the diffused particle has the same frequency
as the incidental one
Incident Photon Scattered Photon
Molecule

In Elastic Scattering the diffused particle has the same frequency as the
incidental one.
Rayleigh scattering is the elastic scattering of light by molecules and particulate
matter much smaller than the wavelength of the incident light. It occurs when light
penetrates gaseous, liquid, or solid phases of matter. It is inversely proportional
to the fourth power of the wavelength of light, which means that the shorter
wavelength in visible white light (violet and blue) are scattered stronger than the
longer wavelengths toward the red end of the visible spectrum. This type of
scattering is therefore responsible for the blue color of the sky during the day and
the orange colors during sunrise and sunset.
Rayleigh scattering is the main cause of signal loss in optical fibers.

35
 2011-01-31

Inelastic Scattering
In Inelastic Scattering the diffused pulse has a different frequency
than the incidental one
Incident Pulses Scattered Pulses
Molecule

Inelastic Scattering is a fundamental scattering process in which the kinetic

energy of an incident particle is not preserved. In this phenomenon some of the
energy of the incident particle is lost or gained: it means that, respectively, the
frequency of the diffused pulse is lower or higher than the frequency of the
incidental one.

36
 2011-01-31

Stimulated Brillouin Scattering

(SBS)
Incidental Photon = Scattered Photon + Phonon

Phonon
Incidental Photon

Scattered Photon

Frequency Shift = = 10-20 GHz

Stimulated Brillouin Scattering (SBS) is an inelastic scattering phenomenon in

which an incident photon can be converted into a scattered photon of slightly
lower energy, usually propagating backwards, and a phonon, a quasi-particle
which describes the vibrations of crystal lattice. The effect can occur
spontaneously even at low optical powers. For higher optical powers, there can
be a stimulated effect, where the optical fields substantially contribute to the
phonon population. Above a certain threshold power of a light beam in a medium,
stimulated Brillouin scattering can reflect most of the power of an incident beam.
Stimulated Brillouin scattering (SBS) is frequently encountered when intense
beams (e.g. laser light) are propagated through a medium such as an optical
fiber.
For silica fibers, the Brillouin frequency shift is of the order of 10–20 GHz.
SBS introduces the most stringent power limit for the amplification and the
passive propagation of narrow-band optical signals in fibers.

37
 2011-01-31

Stimulated Raman Scattering

(SRS)

Frequency Shift = = 10-13 THz

f1f2 I fn f1f2 I fn

NEVERTHELESS

Stimulated Raman Scattering (SRS) is an inelastic scattering

phenomenon present in DWDM systems, where a number of optical
signals is present in the optical fiber at the same time.
When two laser beams with different frequencies propagate together
through a medium, the lower frequency beam can experience optical
amplification at the expense of the higher frequency beam.
In others words this phenomenon leads to a power transfer from higher
frequency signals towards lower frequency ones.
For silica fibers, the Raman frequency shift is of the order of 10–13 THz,
i.e. 100 nm.
Therefore SRS effect results in an unequal power level among the optical
channels at receiver point.
It means that, considering a flat transmission profile, we are expecting to
see, in a frequency scale, a decreasing power profile at receiving side. Or,
in a wavelength scale, an increasing power profile.
Nevertheless

38
 2011-01-31

Raman Amplification

1550 nm
TX RX

RAMAN
PUMP
SIGNAL
POWER
Raman Pump

RAMAN
POWER

RAMAN
TRANFER
AMPLIFIED
SIGNAL ›Contra-directional Raman amplification
›Co-directional Raman amplification

This can be usefully applied in a special application, called Raman amplification,

that is actually based on the same concept, that is power transfer from lower to
higher wavelengths.
To provide amplification to the traffic signals, another signal having a lower
frequency (normally about 100nm lower) is sent backwards into the fiber.
This signal has a high power level and the lowest frequency among those present
in the medium.
This means that there will be a power transfer from it to the traffic channels, so
traffic channels are amplified along the fiber.
Therefore the fiber itself is used as a distributed amplifier.
It must be noticed that Raman amplification can be applied on both sides of the
link, so we can speak of:
•Contra-directional Raman amplification, that is actually the example shown in
this slide, applied to the receiver side,
•Co-directional Raman amplification, when applied to the transmit side
Raman amplification technique is used in case of very long distance DWDM
connections in conjunction with other amplification techniques. Normally this
method allows to cover hundreds of kilometers, but it always introduces a no flat
profile for channel power levels at receiving side of the link, and, as we will see
later on, this is considered an issue to be countered in DWDM systems.

39
 2011-01-31

DWDM Transmission
Systems

Let’s now have a look to a basic DWDM transmission system architecture.

40
 2011-01-31

DWDM transmission system

Bi-directional
transmission

OADM 70-90km OADM

Tx Rx

Amplifier Amplifier Amplifier Amplifier

A DWDM transmission system is basically made up of a series of parallel

transmitters (called TX in this figure), producing the wavelengths to be
transported on the fiber, that are then multiplexed on the fiber itself by means of
Optical Add / Drop Multiplexers (OADM).
On the other end of the link, the aggregated DWDM signal is Demultiplexed again
by means of an OADM and given to the Receiver (RX).
In the middle, as distances to be covered are typically very long and therefore
attenuation pretty high, amplification shall be added to achieve the required
performance.
Amplification is a very important characteristic of DWDM system to be analyzed
in details.

41
 2011-01-31

DWDM Amplifier
Technology

Let’s now introduce Amplification methods used in DWDM systems.

42
 2011-01-31

Erbium Doped Fiber Amplifier (EDFA)

Isolator Coupler Coupler Isolator

Erbium-Doped
Fiber (10–50m)
Pump Pump
Laser Laser

Amplified spontaneous emission (ASE)

The basic Erbium Doped Fibre Amplifier (EDFA) consists of one or more
laser pumps providing extra energy, a section of erbium doped fiber, opto-
isolators to eliminate the noise coming from the laser pumps, and passive
couplers.
Amplification is achieved by stimulated emission of photons from doping
ions in the doped fiber.
The laser pumps excite ions into a higher energy from where they can
decay via stimulated emission of a photon at the signal wavelength back to
a lower energy level.
The output of the EDFA without input signal is called Amplified
Spontaneous Emission (ASE) and this is the most important cause of noise
in this type of amplifiers.

43
 2011-01-31

Amplifier Gain Variation

› Gain variation (GV) = max gain diff. among channels.

GV G
GV

GVmin
C - band λ
Pin [dB] [nm]

The amplification profile in frequency of the EDFA is neither ideal nor flat:
the gain for each channel in DWDM depends on its own frequency.
So we speak of GV, Gain Variation, that is the maximum gain difference
among channels.
By running the amplifiers in the saturation zone, a form of Automatic Gain
Control (AGC) can be realized.
The flatness of the amplifier gain profile depends on the total power at the
amplifier input.
The relation between Input power and GV is called Gain Tilt.
By making the EDFA working in its comfort area the Gain Variation
becomes minimum and the gain profile becomes flatter.

44
 2011-01-31

OSNR Concept

-18 dBm

OSNR = 20 dB

-38 dBm

Amplified spontaneous emission affects OSNR

The most important parameter to evaluate the quality of a DWDM transmission

system is the OSNR, Optical Signal to Noise Ratio, per each transmitted channel:
OSNR is the ratio between the Signal and the Noise power in a small range
around the signal nominal wavelength.
The principal cause for low OSNR is Amplified Spontaneous Emission.
Every time we use an EDFA amplifier we of course increase power, but we
introduce noise as well.
ASE is a residual wide spectrum power coming from the amplification process.
From this point of view a very accurate network planning is required.
In each point of the link the correct amplifier module must be individuated
according to the length of the downstream link.
The right gain for the amplifier must be chosen considering that the higher the
gain the higher the introduced noise.
The only solution against noise is 3R regeneration (reshaping, retiming, re-
amplification), very expensive in DWDM system.
A sufficiently good OSNR value is normally considered to be around 20dB.
If a channel, at the receiving point, has a low OSNR, it can experience a too high
Bit Error Rate (BER).
This is because, at that point, the noise is too strong compared to the Signal.

45
 2011-01-31

Noise Figure and osnr

› NF quantifies the intrinsic “noisiness” of an EDFA

NF @ λN = OSNR OUT /OSNR IN Noise Figure (dB)

10

7.5

5.0

1520 1540 1560 1580

Wavelength (nm)

The lower NF, the better: theoretical limit for an ideal EDFA is 3.0dB

A very important parameter for EDFA applications is the NF, Noise Figure.
NF quantifies the intrinsic “noisiness” of an EDFA and tells us how much the
optical signal to noise ratio of the incoming signal is worsened passing through
the EDFA.
Considering a specific frequency, the NF is calculated as the ratio between
OSNR OUT and OSNR IN.
Therefore a overall diagram can be drawn to show the NF value across all the
bandwidth covered by the EDFA.
The lower NF, the better: theoretical limit for an ideal EDFA is 3.0dB.
In reality the practical NF value is 6dB.

46
 2011-01-31

EDFA TYPES

›BOOSTER ›IN-LINE AMPLIFIERS ›PREAMPLIFIERS

›Typically High gain ›A good combination ›Used in Rx for its very
of input sensitivity good input sensitivity
›Good input sensitivity
and gain

Different kinds of EDFA are available with different characteristics.

Moreover, according to their position and the role in a DWDM link, we can identify
the following families of amplifier:
BOOSTER:
Used to launch the signal, at the fiber optic link input.
Optimized for high output power, its noise figure is not so important.
Works in saturation.
IN-LINE AMPLIFIERS:
Used in between two intermediate spans of fiber, replacing regenerators.
Optimized for both output power and noise figure.
Normally saturated.
PREAMPLIFIERS:
Used at reception side, at the fiber optic link output.
Optimized for high gain and very low noise figure
Normally works in the linear regime.

47
 2011-01-31

equalization
› A DWDM link is properly equalized when all channels are
received with the same power level and the same OSNR.

Attenuation

Stimulated Raman Scattering

Gain Variation and noise (ASE)

As last issue in DWDM transmission systems we have to speak about

equalization.
Let's remember that light propagates along the fiber in different ways according to
its frequency; it means that for each channel in the DWDM spectrum we will
have:
•different attenuation according to the frequencies we are using,
•different behavior in terms of SRS depending on the group of wavelengths and
the power intensity,
•and different amplification due to the fact that, in EDFA amplifier, we will have to
face Gain Variation and Noise.
All these different behaviors bring to a problem, meaning that, in the DWDM
signal, there will be some channel having high power level and high OSNR while
other ones will have low Power and low OSNR.
This situation cannot be tolerated in a DWDM system, therefore we must
introduce a compensation method, called signal equalization.

48
 2011-01-31

Equalization of a link

Tx Rx

Due to the reason we have seen before, from the same starting power level for all
the signal, at the receiver side we will have a different situation for each channel:
some channels can be detected with a good BER, while other channels may have
a low OSNR with high BER, or even cannot be detected at all.

49
 2011-01-31

Equalization of a link

Tx Rx

MONITOR MONITOR
Tx Power Tuning to equalize the link

To overcome all these difficulties, an equalization of the received OSNRs is

necessary to be made during the link setting up, that means doing a Transmitter
Power Tuning to equalize the link.
An Optical Spectrum Analyzer (Monitor), connected to a Receiver monitoring
point, is necessary to equalize the DWDM link.
This instrument gives some information about the power level and the OSNR for
each received channel.
During the equalization procedure the power level for each channel in
transmission and the output power for each amplifier along the link is set in order
to get a flat profile where the instrument is connected to.
This to be sure to have a properly working and at all time stable DWDM link.

50
 2011-01-31

Conclusions
› Main optical characteristics for WDM systems:
› The feasibility of a single channel link has to be evaluated
taking into account following main parameters:
– Attenuation
– OSNR requirements (amplified system)
– Chromatic dispersion and PMD

› Additional effects to be taken into account:

– Non linear effects (not negligible when increasing number of

spans) and derived cross-talk effects
– 2nd order Chromatic Dispersion
– Amplifier gain variation and noise figure
– Equalization

In conclusion, these are the main optical characteristics for WDM systems:
•The feasibility of a single channel link has to be evaluated taking into account
the following main parameters:
• Attenuation
• OSNR requirements (amplified system)
• Chromatic dispersion and PMD.
Moreover other additional effects have to be taken into account:
Non linear effects (not negligible when increasing number of spans) and
•
derived cross-talk effects
• 2nd order Chromatic Dispersion
•Amplifier gain variation and noise figure
•Equalization

51
 2011-01-31

Thank you for taking the time to listen to this WDM Fundamentals course.
For more technical information regarding OTN, and WSON, please view the
relevant Fundamentals courses.

52